REPORT for Stiefel Foundation Small Grant Award, 2009

REPORT for Stiefel Foundation Small Grant Award, 2009


Stiefel Grant Report - CAJohnson

REPORT for Stiefel Foundation Small Grant Award, 2009

PROJECT: Ecology of slave-maker ants and their hosts:

The effect of geographic variation in parasite and host range on co-evolutionary trajectories.

Christine A. Johnson

Division of Invertebrate Zoology, The American Museum of Natural History

Central Park West at 79th Street, New York, NY 10024

(212) 769-5605,

Table of Contents

Report 2

Budget Justification / Itemized Expenditures12


Stiefel Grant Report - CAJohnson

REPORT for Stiefel Foundation Small Grant Award 2009

PROJECT: Ecology of slave-maker ants and their hosts:

The effect of geographic variation in parasite and host range on co-evolutionary trajectories.

Christine Johnson, The American Museum of Natural History, Division of Invertebrate Zoology,

Central Park West @ 79th Str., NY, NY 10035, (212) 769-5605,


Slave-maker ants are specialized social parasites that repeatedly raid colonies of other ant species to acquire brood supplies,which becomethe slave-maker’s work force. Once thought to have little impact on their host, recent work shows varied reciprocal selection strengths between slave-maker social parasites and hostsacross populations. This geographic mosaic of co-evolutionary interaction strengths is likely related to variation in host and parasite range. My two recent studies of an Ohio population with two sympatric slave-makers that share a single host suggest high competition between slave-makers produced ecological shifts in both slave-makers and attenuated the co-evolutionary ‘arms race’ between the evolutionarily older slave-maker and shared host. During 2009, I carried out a laboratory study with slave-makers and their host from the proposed co-evolutionary ‘hot-spot’ Black Rock Forest[1]to test this hypothesis. Here, P. americanus, the prudent slave-maker species in Ohio, has exclusive access to the available host. I found that the demography of NY hosts and slave-makers differed substantially from the Ohio population. Where there is a 1:3:1 ratio of queenless to monogyne (single queen) to polygyne (multiple queen) colonies in Ohio, in BRF there were almost no polygyne colonies.The laboratory arena experiment revealed that P. americanus was indeed more virulent in this population, raiding a significantly greater proportion of brood from host colonies in a shortened time span. The production of slave-maker sexuals (gynes and males) in experimental colonies was also significantly fewer, an indication that co-evolutionary arms race in BRF is strong.


Competition for shared resources is instrumental in structuring the abundance, distribution and resource use of plants and animals (Lack 1954). Most studies of competition focus on paired interactions between enemies (competitors), between targets (prey or host), or between single coevolutionary interactants (enemy and target). Only recently have theoretical and empirical studies begun to examine the interaction between direct (competitor) and indirect (target) effect of competing enemies on each other and on their targets (e.g., Benkman et al. 2001, Johnson & Herbers 2006, Johnson 2008). Enemy and target relationships are typically predator-prey or parasite-host and follow a coevolutionary ‘arms race’ trajectory, whereby virulence and defense levels cycle continually (Dawkins & Krebs 1979). Coevolutionary interactions between a single enemy and target tend to vary throughout a species’ range due to intrinsic and extrinsic effects, and thus produce a geographic mosaic of reciprocal selection strengths (Thompson 1994, 1999, 2005). The outcomes of multipartite coevolutionary interactions are even more variable (e.g., Siddon & Witman 2003) as more enemy options, i.e., the available range (number) of targets (Thompson & Pellmyr 1992) typically temper exposure to selection and multiple enemies force trade-offs by their targets (Sih et al. 1998).

The importance of parasites in evolutionary biology and ecology has long been understood (Price 1980). However, social parasites have traditionally been overlooked as model organisms for studies of competition and co-evolutionary interactions. They occur in such diverse groups as birds, fish, and social insects, and negatively impact their host by parasitizing behaviors such as brood care, feeding and grooming towards themselves (Wilson, 1971) and manipulating sex ratios (Bourke, 1989). Social parasites are often closely related phylogenetically to their hosts and have similar generation times and ecological necessities. Thus, they are more susceptible to cyclical reciprocal selection dynamics than the traditional endo- and ecto-parasites. The avian brood parasites, such as the brown-headed cowbird, have been well studied partly because of their negative impact on songbird populations (Brittingham & Temple, 1983; Robinson et al., 1995). Hymenopteran social parasites are less well understood, but recent work on slave-maker ants dispels past assumptions that their effect on host populations due to their relative rarity is negligible (Hare & Alloway, 2001, Foitzik et al., 2001; Herbers & Foitzik, 2002; Johnson & Herbers, 2006; Johnson, 2008).

Slave-maker ants are specialized social parasites that replenish their host supply by repeatedly raiding colonies of other ant species for their brood. Slave-maker species were once thought to have little impact on their host because of their relative rarity. However, behaviors of host and slave-maker, colony social structure, and sex allocation ratios have been shown to vary among populations, as does the number of interacting species, which indicate intimate co-evolutionary interactions between species (Foitzik & Herbers, 2001; Johnson & Herbers, 2006; Johnson 2008). The overall goal of my research is to examine how variation in host and slave-maker range (i.e., the number of interacting species) across the geographic landscape impacts respective co-evolutionary interaction strengths. The focal taxon is the acorn ants (Myrmicinae: Formicoxenini), which contains several species of slave-makers that nest inside preformed cavities of slightly aged sticks and various nuts found on the forest floor. Two of these slave-makers (Protomognathus [= Harpagoxenus] americanus [Emery], Temnothorax duloticus [Wesson]) are widely distributed in deciduous forests throughout the northeastern United States and southeastern Canada, where they share three closely related host species in the genus Temnothorax[2] in different frequencies (Talbot 1957, Alloway 1980, Alloway et al., 1982). My past research has shown that where these slave-makers are sympatric (co-occur) and share a single host, high competition between slave-makers produced ecological shifts in both slave-makers and attenuated the co-evolutionary ‘arms race’ between the evolutionarily older slave-maker and shared host (Johnson & Herbers, 2006; Johnson 2008): 1) initial raiding times were asynchronous, which afforded a priority advantage to the early raider, 2) intra-specific raiding occurred only among prudent parasite colonies that led to larger single prudent slave-maker colonies that may offer protection against the competing slave-maker species, and 3) there was a higher mutual tolerance of sexuals of both the slave-maker and the host in prudent parasite colonies, which suggests an attenuated arms race for the ‘weaker’ competitor.

During the summer of 2009 with funds from the Stiefel Foundation Small Grant Award, I carried out a study in the putative co-evolutionary ‘hot-spot’, Black Rock Forest, that was designed to test the above hypothesis with the slave-maker P. americanus, which has exclusive access to the available host. First, I found that the demography of NY hosts and slave-makers differs substantially from the Ohio population of hosts and slave-makers. In NY, both host and slave-maker colonies contained fewer individuals than in OH and a higher proportion of the population of host and slave-maker colonies were queenless. Host colonies in Black Rock Forest were also significantly less likely to have multiple host queens. Second, the laboratory arena experiment revealed that the NY P. americanus were more virulent than OH P. americanus. These results suggest that the relatively vast contiguous forest provides ample habitat for these ants and that without a competing slave-maker rival, the slave-maker species that is a prudent parasite in OH exerts strong selective pressure on the exclusive host in Black Rock Forest, indicating an active and strong co-evolutionary arms race.

Study System

The slave-maker ant P. americanus is widely distributed throughout the northeastern U.S. In Black Rock Forest, the slave-maker P. americanus has exclusive access to its host Temnothorax longispinosus. The lack of competitors (Sih et al., 1998) and restricted host use (Kawecki, 1994, 1998) make this site a likely co-evolutionary ‘hot-spot’, where reciprocal selection is strong. In central Ohio, the presence of the competing slave-maker Temnothorax duloticus appears to have shifted the ecological dynamics from those in the eastern population. In OH, T. duloticus raids earlier in the season than P. americanus and consequently, wiped out P. americanus in both laboratory and field experiments and quickly decimated its host before the seasonal production of host sexuals, leaving little opportunity for host recovery. Thus, in OH, there is strong selection pressure on the host to evolve defenses in response to T. duloticus and being parasitized by P. americanus may offer relative benefits. The delayed and attenuated impact on the experimental host population when subjected to both slave-makers along with an inverse relationship in the proportion of the two slave-maker species in natural populationsin OH suggest direct asymmetrical antagonism between slave-makers, the advantage of which belonged to T. duloticus (Johnson & Herbers, 2006). The asymmetrical antagonism between slave-makers appears to have weakened the antagonistic co-eveolutionary relationship betweenP. americanus and the shared host and selected for a more mutualistic relationship compared to the eastern population (Foitzik & Herbers, 2001). The prudent parasite has a less destructive impact on its host and protects future host supplies by rearing host sexuals (Johnson & Herbers, 2006, Johnson, 2008). However, when raiding ‘motivation’ of both slave-makers were synchronized, P. americanus emerged as the better direct competitor, decimating T. duloticus colonies and taking the brood. This indicates that raiding phenology provides T. duloticus the competitive advantage and that reduced virulence in P. americanus is a function of virulent parasite pressure. In NY, there is no antagonism from a competing slave-maker and therefore the dynamics should differ and the co-evolutionary arms race between slave-maker and host should be strong.


Colony collections

During May 2009, colonies of the slave-maker P. americanus and of its primary free-living host in the region, T. longispinosus, were collected from Black Rock Forest, censused and settled between two glass slides (7.5 cm x 2.5 cm) separated by semi-translucent rubber (1.5 mm thickness) that had been carved out to create a small cavity. Slide nests were placed into Petri dishes (8.5 cm diameter), where ants had access to food and water until testing time. From each colony, several ants were preserved in 75% or 95% ethyl alcohol for voucher or future genetic analysis, respectively.

Laboratory Experiment

Experimental design

To assess the impact of parasite and intra-parasite competition on the host, one or two parasite nests were introduced to laboratory arenas (16 cm h x 52 cm l x 35 cm) that contained one queenless and three monogyne (single-queen) nests of T. longispinosus of similar size. Slave-maker nests were placed in the arena center or off-center approximately 4 cm from the walls lengthwise in the paired slave-maker treatment facing outwards. Feeding stations that consisted of a 0.5 cm 2 piece of aluminum foil were placed alongside each nest and replenished with food every other day. Each arena also contained additional glass tube nesting sites (6.5 cm l; 4 mm o.d.) to provide ants the opportunity to relocate. Arena bottoms consisted of Plaster of Paris and sides were coated with Vaseline and double-sided tape to prevent ant escapes. Ants were fed the Bhatkar and Whitcomb (1970) diet and frozen Drosophila.

Treatments were replicated four times for each population for a total of 24 trials (12 for each population). All treatments of a single replicate were activated on a single day, with the four replicates for each population starting within seven days of each other. Arenas were censused daily for the number of queens, workers, and juveniles of each species 1) in each nest to determine the parasite impact on free-living and captive hosts; 2) in the arena and on container sides to account for missing individuals within nests; and 3) in the arena and on container sides clasping another. Mortality counts along with changes in brood number between census dates relative to total brood loss provided a measure of colony raiding in the absence of direct observation.

Trial lengths varied within and among treatments. Trials were terminated when worker numbers from free-living host colonies fell 90%below original numbers. If both slave-maker colonies remained in an arena, the trial was continued until one slave-maker colony raided the other colony. Some trials never met these criteria but were terminated eventually.

Statistical Procedures

Distributions were tested for normality and heteroscedasticity. Data that violated these assumptions were transformed or subjected to non-parametric tests accordingly. Colony demographics were analyzed using repeated measures analysis of variance (RMA) on the proportion of individuals (workers or juveniles) that remained in nests over timeto assess the impact of parasite on parasite or host colonies (workers and brood). The square roots of proportional data were arcsine transformed to normalize the data; ‘arena’ was a random effect and ‘condition’ and time were fixed effects.

Multivariate regressions were used to determine what initial factor (ratio of captive host workers to parasites, sum of captive host workers and parasites, or juveniles in parasite nests) determines the impact on the enclosure host population. Tukey’s HSD (honestly significant differences) and sequential Bonferroni correction (Rice 1989) were used for pairwise comparisons and multiple comparisons, respectively.


Slave-makers were found in 13% of the 192 colonies collected from Black Rock Forest and 5% of the 416 colonies collected from Huyck Preserve during May and June 2009 (Table 1).

Table 1. The number and proportion of slave-maker and host colonies collected from Black Rock Forest (BRF) and Hyuck Preserve (HP).

Slave-maker / Host
BRF / 25 (13%) / 167 (87%)
HP / 20 (5%) / 396 (95%)
BRF / 56 (16%) / 295 (84%)
HP / 22 (8%) / 240 (92%)

Black Rock Forest had fewer queenless colonies of T. longispinosus than Hyuck Preserve, which approximated the proportion of queenless, single queen and multiple queen colonies found in OH, suggesting the relatively large contiguous forest provides ample nesting sites and reduced competition for those sites. There were also fewer queenless slave-maker colonies in BRF (Table 2).

Table 2. Number and proportion of queenless, single queen and multiple queen colonies of the host T. longispinosus and average number of workers in these colonies from Black Rock Forest (BRF) and Hyck Preserve (HP) in 2008 and 2009 and central Ohio 2005-2006.

Queen Number
2008 / 0 / 1 / >1
BRF / N / 165 (49%) / 149 (44%) / 22 (7%)
Mean / 14.78 / 30.77 / 24.45
SD / 10.71 / 34.85 / 30.89
HP / N / 89 (26%) / 162 (46%) / 98 (28%)
Mean / 15.64 / 27.18 / 36.77
SD / 8.83 / 23.50 / 37.36
BRF / N / 102 (62%) / 63 (38%) / 1 (.006%)
Mean / 16.23 / 28.48 / 43
SD / 10.76 / 20.95
HP / N / 106 (27%) / 212 (54%) / 78 (20%)
Mean / 16.11 / 23.97 / 26.96
SD / 9.66 / 16.12 / 17.47
2003 - 2005
Ohio / N / 346 (22%) / 926 (60%) / 282 (18%)
Mean / 26.85 / 41.87 / 56.11
SD / 21.72 / 35.82 / 48.98

Parasite impact on host colonies

Slave-makers had a significant effect on the decline of juveniles in arena host nests (RMA: F2, 94 = 25.13, P < 0.0001) (Figure 1a). The end average loss in arena host colonies with a slave-maker present was 1 1/2 (single slave-maker colony) to 6 two slave-maker colonies) times more than the end loss of host colonies in arenas with no slave-maker present (Tukey’s HSD = 2.39, P = 0.05). The majority of juvenile loss in host colonies occurred within one day after the onset of each trial (RMA: F7, 89 = 9.35, P < 0.0001, Tukey’s HSD = 3.11, P = 0.05).

Slave-makers also had a significant effect on the decline of workers in arena host nests (RMA: F2, 81 = 24.44, p < 0.0001) (Figure 1b). However, a single colony of slave-maker had much smaller effect than two slave-maker colonies (Tukey’s HSD = 2.39, P = 0.05). The majority of juvenile loss in host colonies occurred within the first two days after the onset of each trial (RMA: F7, 89 = 9.35, P < 0.0001, Tukey’s HSD = 3.11, P = 0.05).The impact of two slave-maker colonies on the survival of host colonies queens was also significantly greater than a single or no slave-maker colony (RMA: F2, 23 = 15.77, p < 0.0001, Tukey’s HSD = 2.55, P = 0.05).

Population differences in parasite impact on host colonies

The comparison of parasite impact on host juveniles and host workers between NY and OH revealed that NY slave-makers had a significantly greater impact on host than OH slave-makers, which included the virulent slave-maker Temnothorax duloticus(RMA-juveniles: F1, 143 = 36.3, p < 0.0001; RMA-workers: F1, 143 = 32.19, p < 0.0001) (Figure 2). A between population comparison of juveniles remaining in host colonies on Census Day 6 only with two P. americanus colonies also showed that OH P. americanus had a much greater effect on host colonies (t-test = -6.18, p = 0.0002). Slave-maker impact on host workers on Census Day 6 was not as great (t-test = -2.33, p = 0.052).